as ‘intensity’ see [CLAPHAM and BEARD, 1991])) to 
the same area symbols. In this way the relative 
lightness or darkness of the colours portraying 
the soil classes varies with the reliability of 
the classification (e.g. dark green when a 
particular soil is most reliably classified as 
‘clay’ and light green when a particular soil is 
less reliably classified as ‘clay’ - that is, the 
darker the tint, the more likely a correct 
classification). Such a variation of the lightness 
value of a hue, depending on the (ordered) 
reliability information it represents, can be 
introduced relatively easily in both a hard- and 
soft-copy environment (the only technical problem 
being the relationship between the colour tints as 
they appear on the screen and as they are printed 
on paper). Ordered reliability information can 
better not be shown by overprinting with black dot 
patterns of varying density. The differences in 
value thus created are not wrong in principle, but 
the black dots may make the colour underneath less 
recognizable. 
On the other hand, the combination of visual 
variables in the same set of symbols may sometimes 
lead to unwanted effects on the perception 
properties. Besides, overprinted (open) black dot 
or line patterns of varying density will be needed 
if the visual variable value has already been used 
to represent another aspect of information, for 
instance in a suitability map (the darker the 
colour, the more suitable). It goes without saying 
that there are also limits to the maximum number 
INFORMATION represented by a PERCEPTUAL 
visual variable with PROPERTY 
Quantitative --77---7-777-777777 > Quantitative 
Ordered. 95 3-77 Era > Ordered 
Qualitative ———————————————— > . Associative 
(c. Selective) 
Figure 2 - Essence of the grammar of cartography 
of aspects of information which can be represented 
in the same set of symbols or in the same map. 
Often, some kind of grouping (classification) of 
the information (e.g. on lineage) is needed before 
representation in a single map is possible. 
2.4 Other cartographic ways of dealing with 
quality information and accuracy 
  
  
Next to its representation by means of visual 
variables as applied to symbols, there are also 
other ways in which quality information can be 
reflected in maps. 
For instance, the positional accuracy of soil 
boundaries is often not very high. The generation 
of solid, fine and intricate boundaries in a soil 
map often gives a completely wrong impression of 
the accuracy of these boundaries to the map user. 
Therefore, it can also be considered to completely 
omit the boundaries as line symbols in the map; 
the more or less contrasting colours for the 
different soil units will automatically provide a 
boundary, but a boundary which is less prominent. 
The consequences of cartographic generalization 
(both graphic and conceptual) should also be 
considered carefully. Not only has cartographic 
generalization negative effects on data quality 
(especially on positional and attribute accuracy 
611 
and on completeness), but also there are often 
differences in the levels of generalization of 
cartographic data sets to be integrated. In soil 
mapping, for instance, there are often marked 
differences in the accuracies of the topographic 
base map details and the soil information. 
Cartographic generalization methods may have to be 
applied to adjust the information qualities. 
A final example of a cartographic way of dealing 
with data quality is the so-called dasymetric 
mapping technique, which can be applied in cases 
where (often socio-economic) data are available 
for administrative regions only. These data are 
often represented by choropleth maps in which each 
region receives a uniform tint, suggesting a 
homogeneous distribution of the data over the 
area, which is normally not the case (e.g. think 
of a population density map). With the dasymetric 
mapping technique, the quality of the attribute 
information can be improved by adjusting the 
boundaries of the mapping units to the phenomenon 
represented, with the help of, for instance, 
topographic information (e.g. populations normally 
do not live in swamps, nor in lakes or on the tops 
of high mountains). 
    
  
ORM 
F A P4 I RENTATION 
um © x 
[Red ] 
[Bue | COLOUR 
   
  
  
T : VISUAL y 
POSITION uum 
| 
\ 
e AX A (D 
e ^^ 00 
= TEXTURE 
SIZE c3 
E 
VALUE 
[Green 
Figure 3 - The seven visual variables 
(source: 
BOS, 1984, p.22) 
3. INTEGRATED LAND AND WATERSHED INFORMATION 
MANAGEMENT SYSTEM (ILWIS) 
Turning now to the computer environment in which 
the cartographic ideas presented in the previous 
sections will be investigated, ILWIS was initiated 
some seven years ago at ITC by Meijerink [GORTE et 
al., 1988], where it was developed by the 
Computing Department, for a Watershed Management 
project in Indonesia. It integrates raster 
(particularly satellite), vector (particularly 
cartographic) and tabular data. It is MS DOS PC 
based, but is now being upgraded to run on HP UNIX 
Workstations. Because of its low-cost it has 
rather become an educational ‘workhorse’ at ITC, 
being used (along with other higher-cost systems) 
in several of our postgraduate courses for those 
educational components dealing with image 
processing, ortho-image and ortho-photomapping, 
digital terrain modelling, digital monoplotting, 
on-screen and tablet digitizing, database design, 
and geographic analysis. Furthermore its nature is 
such that researchers (M.Sc. students or staff) 
can implement their own developments as ‘add-ons’ 
to ILWIS. A tradition is emerging at ITC that new 
scientific developments within the institute 
produce an enhancement of ILWIS. (Chaos is 
prevented by a team of professional programmers!)